The Rise of Agentic AI (Capgemini Research Institute)

Introduction
In this post I explore two recent AI research papers directly connected to our course topics — Agentic AI systems and the A* search algorithm.

Paper 1: The Rise of Agentic AI (2025)
This paper reviews how modern AI agents are defined, what frameworks exist for building them, and the key challenges they face. The most interesting finding is that agentic AI systems must handle partial observability, dynamic environments, and multi-agent coordination — exactly the dimensions we studied in class for environment classification.

Connection to Course:
In our AI course we classify environments using 7 dimensions. This paper directly maps to that — agentic AI is designed specifically for environments that are dynamic, stochastic, and partially observable, which matches our GB flood rescue robot scenario perfectly.

Paper 2: A Algorithm Based on Adaptive Weights (2025)*
This paper improves the classic A* search algorithm by dynamically adjusting the heuristic weight based on the current state of the environment. Instead of a fixed f(n) = g(n) + h(n), it uses f(n) = g(n) + w(n)×h(n) where w(n) changes during search.
Connection to Course

We studied A* in class as the best search algorithm for weighted environments. This paper shows how A* can be made even better by adapting its heuristic weight — directly relevant to our rescue robot which operates in a constantly changing flood environment.

Personal Insight
Reading these papers manually first helped me understand the big picture. Using NotebookLM afterwards helped me clarify the technical details I missed, especially the mathematical formulations. The combination of both approaches gave me a much deeper understanding than either alone.

Comments

[Saif Malik]

Wow it's really helpful 😊 thank you 🫶🏼